Enantioselective total synthesis of (+)-milnamide A and evidence of its autoxidation to (+)-milnamide D

Article abstract of DOI:10.1002/anie.200461245

Spontaneous oxidation of the marine-sponge-derived peptide (+)-milnamide A (1) to (+)-milnamide D (2), which is also a natural product, was observed during the total synthesis of 1. The absolute configurations of the structures were determined and are als

Full text of DOI:10.1002/anie.200461245

Angewandte
Chemie
Natural Product Synthesis
Figure 1. Milnamides A, C, and D (1–3), hemiasterlin (4), and hemiasterlins A–C
(5–7). Numbering is taken from Reference [1a].
Enantioselective Total Synthesis of
(+)-Milnamide A and Evidence of Its
Autoxidation to (+)-Milnamide D**
recently, the X-ray crystal structure of 2 was reported,[1c]
however, the absolute stereochemical configurations of 1–3
1
were not defined. Comparison of the H NMR spectroscopic
Chaomin Liu, Makoto N. Masuno, John B. MacMillan,
and Tadeusz F. Molinski*
data of 1–3 suggests that the differences from (ꢀ)-4 are
located solely within the region ofthe heterocyclic amino
acid. Because the highly methylated b-carboline amino acid 8,
which is found in 1–3, has no precedent in other natural
products, the preparation ofboth 3 R and 3S (b-carboline
numbering) stereoisomers of 1 would inform stereochemical
determinations ofthe milnamide afmily. Herein, we report
the first enantioselective total syntheses of (3S)-1 and (3R)-1
that employs an expedient synthesis ofthe requisite 8 through
an oxazoline–dihydrooxazinone rearrangement. The synthe-
sis confirms the absolute configuration of natural (+)-1 as
(3S,12S,15S) and demonstrates the first use of the aforemen-
tioned rearrangement in natural peptide synthesis. Further-
more, we show that 1 undergoes facile autoxidation to
milnamide D (3) which thus correlates the absolute config-
urations ofboth natural product peptides and suggests a
nonbiogenic origin for the iminium salt 3.
Retrosynthetic analysis of 1 (Scheme 1) suggests discon-
nections to a pivotal intermediate, namely the b-carboline
amino acid (8), and the known dipeptide 9.[3a] The precursor
10 to the heterocyclic amino acid arises from aromatic
electrophilic substitution of N-methylindole (11) by an
epoxide 12, which itselfis obtained by a Darzenꢀs-type
condensation of(4 R)-2-chloromethyl-4-phenyloxazoline and
acetone.
Milnamides A,[1a] C,[1b] and D[1c] (1–3), (ꢀ)-hemiasterlin
(4),[2a–d] and hemiasterlins A–C (5–7)[2b,d] (Figure 1) comprise
a small family of exceptionally cytotoxic tripeptides that were
isolated from marine sponges Auletta sp., Hemiasterella
minor, Cymbastella sp. and Siphonochalina sp. Compounds
1–7 are powerful antimitotics that disrupt microtubule
assembly during cell division and inhibit growth ofcultured
tumor cells (e.g. 4, IC50 < 1 nm, MCF-7 breast tumor cells).[4]
Hemiasterlin is a more potent cytotoxin in vitro than
paclitaxel (Taxol) and is comparable to dolastatin 10.[2c] As
a consequence ofthis exceptional activity, a synthetic
analogue of 4, SPA110, has entered advanced phase I clinical
trials for the treatment of solid tumors.[4] X-ray crystallo-
graphic analyses confirmed the relative stereochemistries of 2
and (ꢀ)-4, and by the total synthesis of( ꢀ)-4, the S configu-
rations at each a-amino acid residue were verified.[3,4] Very
[*]C. Liu, M. N. Masuno, J. B. MacMillan, Prof. T. F. Molinski
Department of Chemistry
University of California
Davis, CA 95616 (USA)
Fax: (+1)530-752-8995
E-mail: tfmolinski@ucdavis.edu
The synthesis of 1 is outlined in Scheme 2. The pivotal step
relies upon our SeO2-promoted oxidative rearrangement of2-
alkyl- or 2-(arylmethyl)oxazolines to 3,5-disubstituted dihy-
dro-2H-oxazinones.[5] The required oxazinone 10 was synthe-
sized in two steps as follows. N-methylindole and the epoxide
12[6] were coupled under Lewis acid conditions (SnCl4,
CH2Cl2, ꢀ788C, 56%)[7] to provide the requisite oxazoline
[**]We thank Professor Chris Ireland (University of Utah) for generous
gifts of natural (+)-1 and (+)-3. The Bruker Avance 600 MHz and
Varian Inova 400 MHz spectrometers were purchased with funds
provided by NIH RR11973 and NSF CHE-9808183, respectively. This
work was funded by the NIH (RO1 CA 85602).
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2004, 43, 5951 –5954
DOI: 10.1002/anie.200461245
ꢀ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5951

Communications
Table 1: Oxidative rearrangement of 13 to 10 with SeO2 (2.6 equiv).
Entry
Solvent
T [8C]
t [h]Yield of
13 [%]
[a]
1
2
3
4
5
6
dioxane
dioxane
dioxane
CH2Cl2
EtOAc
100
30
40
40
77
61
2
72
72
72
2[b]
4
–
87
23
53
90
90
CHCl3
[a]Trace. [b]Stirred at 24 8C for an additional 14 h.
phenylglycinol that is used in the preparation ofthe oxazoline
12. Thus, reduction of(5 R)-10 or (5S)-10 would provide (3S)-1
or (3R)-1, respectively. After experimentation under different
conditions (Table 2), we found that 10 underwent hydro-
genation (PtO2, H2, 4 atm, MeOH, RT, 90%) to give 14 with
excellent diastereoselectivity (ꢂ 70:1). The major isomer,
14a, which was separated from the mixture by column
chromatography (silica gel), was shown to have
Scheme 1. Retrosynthetic analysis of 1.
the expected cis configuration by NOE studies;
irradiation ofthe H3 center (CDCl , d = 4.39, s)
3
resulted in an NOE enhancement ofthe benzylic
proton H5 (oxazinone numbering). Condensation
=
of 14a with H2C O under optimized Pictet–
Spengler[10]
conditions
(paraformaldehyde,
MgSO4, benzene, 758C) gave 15 in quantitative
yield. Removal ofthe chiral auxiliary from 15 by
hydrogenolysis–hydrolysis
(Pd(OH)2/C,
H2,
4 atm, TFA/H2O, 89%),[11] followed by reductive
methylation (paraformaldehyde, Pd/C, H2,
MeOH, 508C, 78%)[12] provided the key amino
acid 8.
The amide coupling of 8 and 9[3a] was far more
troublesome than expected. Attempted amide
bond formation with
a variety of coupling
reagents (dicyclohexyl carbodiimide (DCC),
1-ethyl-3-(3-dimethylaminopropyl)carbodiimide)
hydrochloride (EDC), O-(7-azabenzotriazol-1-yl-
N,N,N’,N’-tetramethyluronium hexafluorophos-
phate (HATU), bromotripyrrolidinophospho-
nium hexafluorophosphate (PyBrOP)) all failed
to produce the expected tripeptide. Clearly, the
highly hindered molecule 8 precluded the forma-
tion ofthe amide bond with the sterically crowded
tert-leucyl amino terminus of 8 under conditions
Scheme 2. Synthesis of (+)-milnamide A (1): a) 12,[6] SnCl4, CH2Cl2, ꢀ788C (56%);
b) SeO2, CHCl3, reflux (90%); c) PtO2, H2, MeOH, room temperature (90%); d) para-
formaldehyde, MgSO4, benzene, 758C (100%); e) Pd(OH)2/C, H2, 4 atm, TFA, H2O
room temperature (89%); f) paraformaldehyde, Pd/C (10%), H2, MeOH, 508C
(78%); g) 9,[3a] pivaloyl chloride, iPr2EtN, THF, 08C (10%); h) LiOH, MeOH, H2O,
degassed, room temperature (92%). TFA=trifluoroacetic acid.
13 as a mixture ofepimers. [8] Attempted oxidative rearrange-
ment of 13 to the required 5,6-dihydro-2H-1,4-oxazin-3-one
10 with SeO2 under standard conditions (dioxane, reflux) was
that were similar to those employed by Andersen et al in the
synthesis of 4 from the less hindered (2S)-N-Boc-N,N’C,C-
tetramethyltryptophan (52%).[3a] Eventually, amide coupling
was possible under mixed anhydride conditions (pivaloyl
chloride, iPr2EtN, THF, 08C, 10%) to give the ethyl ester
17.[13] Saponification of 17 (LiOH, MeOH, H2O, degassed, N2,
disappointing (ꢁ 23% yield), probably owing to decomposi-
[9]
tion ofthe indole ring.
A survey ofvarying conditions
(Table 1) identified that SeO2 in CHCl3 or ethyl acetate at
reflux provided rapid reactions and the highest yields.
Accordingly, oxazoline 13 was smoothly converted (SeO2,
CHCl3, reflux) into 10 in 90% yield.
Table 2: Hydrogenation of 10 in the presence of PtO2 (20 mol%).
Entry Solvent PH [atm] t [h]d.r. [b] (14a:14b) Yield of 14a,b [%]
2
Control ofthe correct configuration at the C3 center in
8
1
2
3
CH2Cl2
EtOAc
CH3OH
2
4
4
48
52
72
17:1
39:1
72:1
90
was anticipated from the diastereofacial hydrogenation of the
96
=
C N bond directed by the Ph substituent at the C5 center of
90[a]
the dihydrooxazinone 10. The synthesis ofeither enantiomer
of 8 is possible by the choice ofthe conifguration ofthe
[a]Isolated yield of 14a. [b]From integration of 1H NMR spectra.
5952
ꢀ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org Angew. Chem. Int. Ed. 2004, 43, 5951 –5954

Angewandte
Chemie
RT, 92%) completed the synthesis of(3 S)-1. Its counterpart
(3R)-1[14] was synthesized by the same route starting with the
(4S)-oxazoline 12. Synthetic (3S)-1 was identical to natural
(+)-1 from 1H NMR and circular dichroism (CD; see
Figure 2) spectroscopic analysis, ESI-MS measurements, and
LC–MS retention times.[15] However, the CD spectrum ofepi-
milnamide A ((3R)-1) differed significantly from that of
natural (+)-1.
Keywords: carboline · enantioselectivity · natural products ·
peptides · total synthesis
.
[1] a) P. Crews, J. J. Farias, R. Emrich, P. A. Keifer, J. Org. Chem.
1994, 59, 2932; b) R. N. Sonnenschein, J. J. Farias, K. Tenney,
S. L. Mooberry, E. Lobkovsky, J. Clardy, P. Crews, Org. Lett.
2004, 6, 779; c) C. Chevallier, A. D. Richardson, M. C. Edler, E.
Hamel, M. K. Harper, C. M. Ireland, Org. Lett. 2003, 5, 3737.
[2] a) R. Talpir, Y. Benayahu, Y.
Kashman, L. Pannell, M. Schleyer,
Tetrahedron Lett. 1994, 35, 4453;
b) J. E. Coleman, E. D. de Silva, F.
Kong, R. J. Anderson, T. M. Allen,
Tetrahedron 1995, 51, 10653;
c) H. J. Anderson, J. E. Coleman,
R. J. Andersen, M. Roberge,
Cancer Chemother. Pharmacol.
1997, 39, 223; d) Y. Kashman,
D. G. Gravalos,
US Patent
5661175, 1997; e) W. R. Gamble,
N. A. Durso, R. W. Fuller, C. K.
Westergaard, T. R. Johnson, D. L.
Sackett, E. Hamel, J. H. Cardelli-
na II, M. R. Boyd, Bioorg. Med.
Chem. 1999, 7, 1611.
[3] a) R. J. Andersen, J. E. Coleman,
E. Piers, D. J. Wallace, Tetrahedron
Lett. 1997, 38, 317; b) E. Vedejs, C.
Kongkittingam, J. Org. Chem.
2001, 66, 7355.
[4] J. A. Nieman, J. E. Coleman, D. J.
Wallace, E. Piers, L. Y. Lim, M.
Roberge, R. J. Anderson, J. Nat.
Prod. 2003, 66, 183.
[5] C. M. Shafer, T. F. Molinski, J.
Org. Chem. 1996, 61, 2044.
[6] a) K. Kamato, H. Sato, E. Takagi,
I. Agata, A. I. Meyers, Heterocy-
cles 1999, 51, 373; b) S. Flori, V.
Figure 2. CD spectra of milnamides: a) synthetic milnamide A [(3S)-1] (c), synthetic epi-
milnamide A [(3R)-1] (b), and natural milnamide A [(3S)-1] (a); b) synthetic milnamide D [(3S)-3]
(b) and natural (+)-milnamide D [(3S)-3] (c).
Over time, 1 slowly autoxidized to 3 upon standing in
Capriati, R. Luisi, Tetrahedron Lett. 1996, 37, 4781.
solvent. Both (3S)-1 and (3R)-1 were oxidized at similar rates,
but the apparent rate ofautoxidation was greatly accelerated
when samples were stored in aged CDCl3 or CHCl3.[16] As the
analyses ofthis autoxidation product 3 were identical to those
ofnatural milnamide D by CD (Figure 2) and 1H NMR
spectroscopy and LCMS, we conclude that the latter also has
the 3S configuration. Given the ease of this oxidation, it is
plausible that 3, which is obtained from natural sources,
originates from the autoxidation of 1 during its isolation–
purification process, although we cannot exclude 3 as a “true”
natural product or an intermediate precursor in the biosyn-
thesis of 1.
The key feature ofthis synthesis of( +)-milnamide A (1) is
the high-yielding preparation ofthe highly methylated b-
carboline amino acid 8, which is made possible through the
facile oxidative rearrangement of oxazoline 12 to the
corresponding substituted dihydrooxazinone. This rearrange-
ment reaction was exploited for the first time in natural
product synthesis for the preparation of amino acid 8 and
should find application in the synthesis of other marine-
derived peptides that containing rare tert-alkyl amino acids.[17]
[7] R. Reddy, J. B. Jaquith, V. R. Neelagiri, S. Saleh-Hanna, T.
Durst, Org. Lett. 2002, 4, 695.
[8] All new compounds gave satisfactory HRMS and 1H and
13C NMR spectroscopic data. The resulting 1:1 epimeric mixture
of 13 was inconsequential as both epimers were converted into
10 in the subsequent reaction.
[9] We assume from the slower rate of the SeO2 oxidation reaction
under the original conditions (reference [5]; anhydrous 1,4-
dioxane, reflux) that oxidative degradation of the indole ring is
competitive with the oxazoline!oxazinone rearrangement.
[10] Addition ofstoichiometric or catalytic quantities ofBrønsted
acid (p-toluenesulfonic acid or CF3COOH) to the reaction led to
decomposition ofthe substrate.
[11] L. M. Harwood, S. N. G. Tyler, A. S. Anslow, I. D. MacGlip,
M. G. B. Drew, Tetrahedron: Asymmetry 1997, 8, 4007.
[12] P. L. Ornstein, D. D. Schoepp, M. B. Arnold, N. K. Augenstein,
D. Lodge, J. D. Millar, J. Chambers, J. Campbell, J. W. Paschal,
D. M. Zimmerman, J. D. Leander, J. Med. Chem. 1992, 35, 3547.
[13] Ester 17 was purified by HPLC (C18 column; gradient, CH3CN/
H2O with HCOOH (0.01%)). A by-product, N-pivaloyl-9, which
was observed during the coupling reaction, attests to the
presence ofthe highly congested carboxy group in the mixed
pivalic anhydride of 8.
[14] Significant differences between the (3R)-1 epimer and (3S)-(+)-
1 were seen in both the 1H NMR and the CD spectra (see
Experimental and Supporting Information).
Received: July 9, 2004
Angew. Chem. Int. Ed. 2004, 43, 5951 –5954
www.angewandte.org
ꢀ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5953

Communications
1
[15] CD, UV, LCMS, and H NMR spectroscopic data for synthetic
(3S)-1 and natural (+)-1 were identical. The HPLC retention
time (co-injection) were also identical.
[16] The ethyl ester, 17, was also sensitive to autoxidation. Handling
ofNMR samples of 1 in 100% CDCl3 from freshly opened
ampoules did not induce autoxidation. Traces ofchlorine or
phosgene, which are present in aged CHCl3, are possible
initiators ofthe autoxidation of 1. The iminium salt 3 is easily
reduced back to 1 (NaBH4, MeOH, 100%).
[17] T. Hamada, T. Sugawera, S. Matsunaga, N. Fusetani, Tetrahedron
Lett. 1994, 35, 609.
5954
ꢀ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2004, 43, 5951 –5954